What the survey shows — and what it does not prove
In the McKinsey survey, 80% of respondents using AI in their work report improved individual productivity. A positive effect on their organization’s earnings before interest and taxes (EBIT) is reported by 37% of respondents. These are different measures, not stages in a single company success funnel.
About 6% of respondents are classified as AI high performers: they attribute at least 5% of organizational EBIT to AI and report significant value. This does not mean that all other organizations receive no benefits.
The survey ran from May 4 to June 8, 2026, with 1,719 respondents in 97 countries. These are respondents’ assessments, not audited financial results. It is not a survey of Lithuanian companies and does not validate the LeanAI method. Source: report pages 3, 11–12, 17 and 30.
Start with the constraint on business results
Our interpretation: before choosing an AI tool, identify what most limits growth, delivery reliability or profitability. It may be delays, long order lead times, quality issues, high costs or cash tied up in inventory.
For example, a faster sales proposal will not remove a production capacity constraint. A faster report will not improve delivery reliability if it does not inform decisions. These are illustrative examples, not LeanAI client results.
The first question is more specific than “where can we use AI?”: “which business result do we want to change, and what currently limits it?”
Redesign the process and prepare people for a new way of working
Respondents reporting higher AI results more often describe fundamental workflow redesign, leadership commitment and impact measurement. The survey identifies associations with reported results; it does not establish causation. Source: pages 18–19.
From the LeanAI perspective, change involves more than automating a task. It requires understanding the process, data quality, handoffs and responsibility for the new way of working.
Agree in advance when a person must validate AI output, what the system may do independently and when an action must stop. Teams need practical skills and clear working standards.
Assess the overall impact and cost
The baseline should reflect the chosen problem: on-time delivery, process lead time, defects or rework, unit cost, inventory turnover or another relevant business measure. AI usage frequency alone does not demonstrate results.
Assess licenses, integration, data preparation, training, human validation and maintenance. Evaluate time savings alongside how the released capacity will be used to benefit the business.
A focused pilot helps test the selected solution. Scale it when agreed measures show useful improvement and the costs and risks are acceptable.
Five questions before starting an AI initiative
1. Which business challenge matters most now, and what measure reflects it?
2. Which part of the process most limits that result?
3. How will the process, people’s work and responsibilities change?
4. What are the full costs, data requirements and human validation needs?
5. How will we decide whether to scale, adjust or stop the solution?
How this relates to IMPACT
IMPACT is our working sequence: Identify → Map → Prioritize → Apply → Codify → Tune. Start with the business priority, investigate causes, select an initiative, test the change, establish the new way of working and review the result.
The McKinsey survey provides broader context. Our practical response is to connect Lean principles and suitable AI capabilities to the organization’s specific situation. We begin by discussing that situation with the leader and choose the next direction based on the priorities identified.
Which challenge most limits your business results?
The first step is a meeting with the leader to understand the current situation and expectations.
Let’s discuss your business challenges →IMPACT method →